What problem does it solve? After code changes are made, teams often ship inconsistent documentation, unmet plan requirements, and speculative dead code left behind by AI assistants. This Skill runs a structured three-layer audit that catches requirement gaps, code defects, and documentation drift before merge. ## Core Features & Use Cases - Three-layer review engine: Layer A audits plan/requirement compliance, DI injection patterns, and documentation sync; Layer B uses Alibaba OpenCodeReview (OCR) delegate mode for deterministic diff scoping and rule-based file review; Layer C reconciles, deduplicates, and re-rates findings into BLOCKER/MAJOR/MINOR severities. - Documentation synchronization checklist: Verifies CHANGELOG, DEVELOPER_GUIDE, README, API docs, i18n locales (zh/en/ru), code comments, config annotations, plan files, and frontend versions.json stay consistent with the change. - AI dead code cleanup: Detects and removes speculative abstractions, single-implementation interfaces, never-populated fields, and unreferenced methods with repo-wide zero-reference scans and safe-deletion boundaries. - Use Case: After finishing a feature branch that touches the API, frontend, and export columns, run this Skill to confirm every plan requirement is implemented, review the diff against OCR rules, fix blockers, and verify all docs and locale files were updated before opening a PR. ## Quick Start Ask the agent to run the review-with-docs skill on the current branch changes to audit plan compliance, code quality, dead code, and documentation synchronization.
